Professional Documents
Culture Documents
Abstract
Differential foundation settlement can impose significant demands on a buildings seismic force resisting system and reduce
the structures seismic capability. This paper describes an inventive evaluation methodology of the effects of such
settlement on a tall reinforced concrete shear wall building. Seismic evaluation of building structures experiencing large
differential foundation settlement is a complex task given current limitations of commercially-available analysis software,
as well as many geotechnical and structural uncertainties. Under static conditions, the effect of differential settlement on
structural adequacy can be assessed by modeling the structure and its foundation, imposing a deformed shape on the
foundation model or base supports to match measured or estimated settlement contours, and evaluating structural demands
on the structure and foundation system. However, for seismic analysis, the foundation deformation should adjust as the
mathematical representation of the building structure is subjected to simulated ground motions and as mat nonlinearity
develops. The evaluation of the high-rise building and its foundation was carried out in several stages using CSI Perform-
3D. The pile cap and mat foundation supporting the building were represented by a nonlinear beam grillage model with
flexural and shear hinges, which was validated through finite element analysis. Hinge properties were estimated based on
multiple layers of top and bottom flexural reinforcement, as well as vertical dowels used as shear reinforcement, provided
throughout the foundation. Externally applied vertical point loads in select locations of the mat and underlying vertical soil
spring properties were iteratively determined to produce a settlement profile matching measured deformations, while
allowing additional mat deformation under ground motion excitation. The building model included explicit representation of
basement retaining walls, first floor slab, building core walls, moment resisting frames, and outriggers to capture the relative
stiffness of these elements and their effect on the foundation deformation. A suite of 7 ground motion records was imposed
on the combined superstructure-foundation model to assess the effect of settlements on the buildings seismic adequacy
using Performance-Based acceptance criteria.
Keywords: nonlinear grillage; differential settlements; mat foundation; pile cap; seismic evaluation
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
1. Introduction
Seismic evaluation of building structures experiencing large foundation settlement is a complex task given
current limitations of commercially-available analysis software, as well as many geotechnical, structural and
modeling uncertainties. To assess differential settlement-related demands on the superstructure, typical
engineering practice involves imposing a deformed shape on the mat or base supports for the structure to match
measured or estimated settlement contours. However, this modeling approach has several shortcomings. First,
the foundation deformations should adjust as the mathematical representation of the building structure is
subjected to simulated ground motions. Second, as nonlinear behavior develops in the mat due to anticipated
flexural or shear failures, the deformed mat shape should adjust accordingly to capture nonlinear effects. Finally,
settlement contours are developed based on a finite number of settlement measurement points, which typically
neglect the stiffness and nonlinear response of the building superstructure, basement walls, or soil-structure
interaction. Thus, strictly imposing deformation contours on a building mat foundation while disregarding these
effects can lead to severe inaccuracies in force demand estimates on the supporting mat, as well as connecting
superstructure. This paper presents the details of an inventive modeling approach that overcomes the above-
mentioned difficulties. This methodology was used for nonlinear modeling and seismic performance assessment
of a pile-supported high-rise building that underwent differential settlement.
To evaluate anticipated building behavior under seismic demands, a nonlinear computer model of the
building and its foundation was developed using CSI Perform-3D V5.0.0 (Perform) [1]. The building is
supported on a continuous, thick, cast-in-place, reinforced concrete pile cap connecting evenly-spaced precast
concrete piles. Initial mat evaluation was carried out using an elastic shell model with effective cracked cross
section properties. Imposed deformations on this model coinciding with a finite number of measured settlement
locations were obtained by specifying support spring loads. The model included explicit representation of
basement retaining walls, first floor slab, building core walls and outrigger frames to capture the relative
stiffness of these elements and their effect on the mats deformation contours.
A second modeling stage involved removal of the support springs and replacement with externally applied
vertical point loads on the mat and underlying vertical soil springs to allow mat deformation under ground
motion excitation. Since net tensile demands were not anticipated on the piles in the current assessment due to
high gravity loads, compression-only soil springs were defined throughout the mat. An iterative process
involving adjustment of the distributed soil spring stiffness at different areas of the mat and the magnitude of
external loads applied at select locations was used to match the smooth deformed shape under gravity loads and
differential settlements previously calculated.
A third modeling stage consisted of replacing the elastic shell elements with a nonlinear beam grillage that
includes flexural and shear hinges and which was incorporated into a global model of the building. Hinge
properties were estimated based on the multiple layers of top and bottom flexural reinforcement, as well as
vertical dowels used as shear reinforcement, provided throughout the mat. Grillage beam model assumptions are
validated through finite element analysis. A suite of 7 ground motion records was imposed to assess the effect of
settlements on the buildings seismic adequacy using Performance-Based acceptance criteria.
2
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
slabs at the bottom two levels, diaphragm constraints in remaining levels, and several representations of the pile
cap, discussed below.
3
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
(a) (b)
Fig. 2 (a) Pile cap foundation plan; (b) Typical pile cap cross section and reinforcement details.
4
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
N N
3. Foundation Model
3.1 Elastic Shell Model
The initial modeling of the foundation employed elastic shell elements in Perform. The thick pile cap and mat
foundation were assumed to be uncracked and unreduced gross section properties were used. This relatively
simple model was meshed to accommodate shear wall and column layout, as well as settlement measurement
points. Figure 4(a) presents a plan view of this model. As can be seen, mesh size was refined near locations of
superstructure vertical elements. Due to Perform modeling limitations, mesh regularity was drastically distorted
to match shear wall and column layout as well as the irregular location of the 31 measurement points. Fig. 4(b)
presents the nonlinear grillage model used in later analyses. The elastic shell foundation model was used for
preliminary assessment of differential settlement effects on superstructure seismic performance. This assessment
included preliminary demand-to-capacity ratios and degree of nonlinear action in the superstructure and
foundation system due to settlement and ground motion. Although these results may be highly inaccurate due to
force redistribution and nonlinearity, this initial modeling stage allowed for quick determination of problem
areas and possible seismic deficiencies.
5
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
N N
In the preliminary study, applied differential settlements consisted of imposed vertical displacement at
each pile cap node using linearly interpolated values between mapped settlement contours. The pile cap
deformation remained constant throughout the response history analysis. This model did not account for the
relative stiffness and nonlinear response of the superstructure and foundation, nor the effect of ground motion
excitation on pile cap deformation. Analysis results for the settlement only case were found to be inconsistent
with the building condition, predicting severe shear and flexural failures in areas that do not have any observed
cracking or distress. Demand-to-capacity ratios were also highly sensitive to very small variations of applied
settlements and pile cap meshing.
Given inconsistencies between analysis results and physical observations, a revised analysis was
performed in which the differential settlements were applied only at the 31 measurement points. Differential
settlements were imposed by defining very stiff support springs at the 31 locations. Support spring element loads
were applied as vertical deformations, where a negative sign represents a downward settlement while a positive
sign represents an upward displacement. In this analysis, no other soil springs were present. Since only the 31
measurement points were restrained, the rest of the pile cap was free to deform, constrained only by the stiffness
of the foundation system and superstructure. Pile cap corners were restrained against lateral displacements. Fig.
5 shows the resulting pile cap deformation. Accuracy of this modeling approach was limited by the essentially
rigid supports at the 31 measurement points.
6
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
(-2.4 in)
(-2.4 in)
(-3.3 in) (-0.3 in)
Fig. 6 Deformed shape of elastic grillage model with EI eff =0.6EI gross (amplification factor=100)
The selected grillage beam spacing resulted in very deep rectangular beams measuring 60 in width (equal
to beam spacing at 5 ft on center) by 120 in deep (corresponding to a 10 ft thick slab). The adequacy of this
modeling approach was validated by comparing the flexural and shear force-deformation relationship of simple
elastic beam models in Perform to thick finite element solid models in Abaqus v6.13.1 (see Table 1).
7
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
V=1200 kip
V=300 kip
M=70,000 kip-in
=1.432e-4 rad
=2.673e-4 rad 1%
M=70,000 kip-in
=2.634e-4 rad
Two beam lengths were assessed: 60 in. (5 ft. beam spacing in orthogonal direction) and 240 in. (4 beams
in series arbitrarily selected to represent a longer beam span). In the Abaqus beam/slab model a very fine mesh
size of 5x5x5 in. of solid 3D deformable elements with 8 integration points were used. Table 1 shows the loads
and boundary conditions used in both Perform and Abaqus models. Comparison of analysis results displays
small differences in shear and flexural force-deformation relationships, thus demonstrating the adequacy of this
modeling approach.
8
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
N N
(a) Grillage model with vertical springs (b) rigid beam elements
Fig. 7 Grillage model with vertical soil springs and stiffened beams (highlighted in red)
The preliminary use of constant soil spring stiffness and an initial set of settlement loads did not result in
the desired deformed shape. To approximate this shape it was necessary to iteratively adjust the spring stiffness
and applied loads, initially using elastic grillage beam elements, then a second time, after adding nonlinear
grillage beam hinges, and finally after adding the basement retaining walls around the pile cap perimeter and an
explicit representation of the first floor slab. Throughout these iterations the soil spring values were reduced
where high settlements were recorded and increased where measured differential settlements were relatively low.
The final soil spring values ranged from 170-1550 kip/in, which correspond to a range of subgrade modulus, k s
of 50-430 pci, with an average k s,base of 200 pci, as shown in Fig. 8. Similarly, the applied settlement loads were
scaled to achieve the desired level of deformation. The dead load and expected live load (25% of the unreduced
live load) in the entire superstructure were included in these iterations to obtain the final deformed settlement
shape. The resulting deformed shape, shown in Fig. 9 matched within 10% accuracy the settlements at the 31
measurement points and approximately matches the settlement contours of Fig. 3.
9
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
(-2.4 in)
(-3.3 in) (-0.3 in)
Fig. 9 Deformed shape of final nonlinear grillage model with distributed soil springs, retaining walls and first
floor slab (amplification factor=100).
10
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
The FEMA Beam, Concrete Type element was used to model nonlinear flexural behavior of the grillage.
Nonlinear flexural parameters were defined per Table 6-7 of ASCE 41-06 [5], assuming a relatively symmetrical
flexural reinforcement layout and high shear demands. To achieve convergence of the settlement loading and
deformed shape, strength degradation was not included in the material model. Instead, the performance of the
beams was monitored using a limiting plastic hinge rotation. The positive and negative flexural capacities of the
pile cap were calculated (as beam segments measuring 5 ft wide) throughout the floor area according to
reinforcement details in Fig. 2 using expected material properties for concrete and reinforcing. Three main areas
of the pile cap and mat foundation were identified, i.e., around the core and outriggers (where high flexural
reinforcement ratios are used), the remaining areas of the 10 ft thick slab (where intermediate reinforcement
ratios are used), and the 3 ft thick mat (with a relatively low reinforcement ratio).
Table 2 Seismic performance criteria defined for superstructure and foundation system
Element Deformation Limit (CP)
Building global behavior Story drift, /h (%) 3.0% average
Concrete compressive strain, cu,Compr 0.005 (in/in)
Reinforcing steel compressive strain, su,Compr 0.02 (in/in)
Core shear wall
Reinforcing steel tensile strain, su,Tens 0.05 (in/in)
Shear strain (drift ratio, /h) 1.0%
Outrigger coupling beams Shear strain (drift ratio, /h) 6.0%
Outrigger concrete spandrel beams Plastic hinge rotation, pl 4.0%
Reinforced concrete frame beams Plastic hinge rotation, pl 4.0%
Reinforced concrete columns Plastic hinge rotation, pl 0.8-1.8%
Embedded steel coupling beams Shear hinge displacement, = p L p =3.0%
Slab plastic rotation at point of strength loss, pl 1.0%
Pile cap foundation
Slab plastic shear hinge displacement, = p L p =0%
Superstructure seismic performance criteria corresponding to Collapse Prevention (CP) level was defined
in accordance with PEERs Tall Building Initiative [2] and ASCE 41-06 [5] for different elements and material
types (see Table 2 above). For the foundation system, nonlinear flexural and shear performance criteria was
taken from the ASCE 41-06 [5] limits for reinforced concrete beams, also shown in Table 2.
11
16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017
Usage ratios (ratios between maximum deformation and limiting deformation criteria) were computed for
all 7 ground motion records for each element in the model. The compliance with performance limits (i.e., usage
ratios smaller than 1), as well as the percent increase in maximum usage ratios due to differential settlements
were determined for all elements. The assessment results are not within this papers scope. However, this
methodology allows identification of requirements for local or global strengthening and seismic retrofit of the
high-rise superstructure elements.
5. Conclusions
This paper presents an inventive methodology that was successfully used to assess the effect of large differential
settlements on a high rise buildings seismic capability. The main difficulty in this assessment was imposing an
initial state of settlements on the foundation system while allowing it to deform freely under ground motion
excitation of the building. Another obstacle was incorporation of the stiffness and nonlinearity of the
superstructure, foundation system, and underlying soil support.
In this study, the common yet extremely limiting use of elastic shell elements to model foundation slabs
was replaced with a versatile grillage beam model, which assumptions were validated through finite element
analysis. Severe elastic shell element distortion was resolved through use of a regular grillage beam layout
defined with nonlinear shear and flexural hinges. The elastic stiffness of the grillage beams was calibrated to
match that of a thick mat or pile cap foundation. Nonlinear vertical soil springs with asymmetric compressive
and tensile behavior were defined underneath each node of the grillage model, with a preliminary uniform elastic
stiffness value throughout the foundation.
Rather than imposing a constant deformed shape on the foundation slab to exactly match incorrect
settlement contours that ignore the superstructure stiffness and may therefore result in significant over or
underestimation of superstructure and foundation demands, a compound iterative process was developed. Initial
differential settlements were applied using vertical loads on select points of the grillage model corresponding to
the locations of settlement measurements at the building site. The magnitude of these loads as well as the soil
spring stiffness distribution throughout the foundation were determined iteratively to closely match measured
settlement values and generally approximate the overall deformed shape of the foundation slab to settlement
contours. As the building model was subjected to ground motion simulations using response history analysis, the
methodology allowed the mat to deform while accounting for the relative stiffness and nonlinearity of the
superstructure and foundation system, as well as soil-structure interaction.
6. References
[1] CSI (2011): User Guide Perform 3DTM Nonlinear Analysis and Performance Assessment for 3D Structures,
Version 5, Computers & Structures, Inc., Berkeley, California, USA.
[2] PEER (2010): Tall Buildings Initiative Guidelines for Performance-Based Seismic Design of Building
Structures, Version 1.0. Technical Report No. 2010/05, Pacific Earthquake Engineering Research Center,
Berkeley, California, USA.
[3] Terzaghi, K. (1955): Evaluation of coefficients of subgrade reaction. Gotechnique 4, 297-326.
[4] ACI (2011): Building Code Requirements for Structural Concrete (ACI 318-11) and Commentary, American
Concrete Institute Committee 318.
[5] ASCE (2006): Seismic Rehabilitation of Existing Buildings, ASCE/SEI Standard 41-06, American Society of Civil
Engineers, Reston, Virginia, USA.
12